372 L.M. Shelton J. of Economic Behavior Org. 41 2000 363–383
depresses target firm gains due to the possible transfer of wealth to bondholders. Therefore, a variable indicating equity financing should be negative in target and bidder equations.
3. Method
3.1. Data The acquisitions in this study were obtained by randomly selecting bidding firms accord-
ing to the methodology of Rumelt 1974, 1978. Rumelt collected his sample by randomly selecting 100 Fortune 500 industrial companies in 1949, 1959, and 1969 and 50 of these
companies in 1974. An additional random sample was taken of 100 Fortune 500 industrial companies in 1979 to include more mergers that occurred in the late 1970’s and early 1980’s.
These randomly selected firms made 199 acquisitions during 1962–1983 that possessed the following characteristics:
1. both target and bidder appear on the CRSP tapes; 2. sufficient line of business data is available for both target and bidder to determine the fol-
lowing information for each business unit: percentage of corporate revenue contributed, the products sold, and customers served;
3. sufficient information on institutional investor holdings is available from the SP Stock Guide.
The necessary line of business data to classify relatedness were obtained through annual reports, prospectuses and Moody’s Industrial Manual.
As Table 1 shows, the study time period includes 11 years of peak merger activity — 1962–1969, 1981–1983, and 11 years of slow merger activity — 1970–1980. The merger
peak and trough years identified by Golbe and White 1993 in their longitudinal study of merger wave activity from the late 1800’s are highlighted in boldface. A total of 46.2 percent
or 92 of the mergers occurred during peak periods, and 53.8 percent occurred during trough periods. The fairly equal distribution of the sample across peak and trough years permits
effective testing of the differences in the two types of mergers.
3.2. Dependent variables The dependent variables, the percent change in bidder equity and the percent change in
target equity were derived using event study methodology to estimate the change in the bidder and target firm share prices due to merger. First, the expected share price without
merger was estimated using the market model employed by Dodd 1980:
R
j t
+ α
j
+ β
j
× R
mt
+ ǫ
j t
where R
j t
is rate of return on stock j over period t, t = 1 day; R
mt
is rate of return on value weighted market portfolio over period t; α
j
= ER
j t
−β
j
× ER
mt
; ǫ
j t
is disturbance term of security j in period t, Eǫ
j t
= 0; β
j
= cov R
j t
, R
mt
var R
mt
. For each merger, α
j
and β
j
were estimated for both the acquiring and acquired firms for a period of 250 trading days ending approximately 3 months before the merger press date.
L.M. Shelton J. of Economic Behavior Org. 41 2000 363–383 373
Table 1 Merger peaks and troughs during 1962–1983 transactions valued at 1 million or more
Year Number of mergers
Peak years Trough years
1962 1047
a
1963 1102
1964 1161
11 11
1965 1222
1966 1286
1967 1354
1962–1969 1970–1980
1968 1829
1969 1712
−23 1981–1983
1970 1318
1971 1269
1972 1263
1973 1064
1974 926
1975 981
1976 1145
1977 1209
1978 1452
1979 1527
1980 1558
48 1981
2328 1982
2298 1983
2393
a
1962–1966 extrapolated from Merger and Acquisitions and Federal Trade Commission data; 1967–1983 Mergers and Acquisitions
data. Sources: Mergers and Acquisitions, JanuaryFebruary 1986, MayJune 1990; Golbe and White, 1993, Review of Economics and Statistics, 493–499, US Federal Trade Commission Bureau of
Economics, 1979, Statistical Report on Mergers and Acquisitions.
This established base values for α
j
and β
j
to calculate R
j t
without merger effects. Over 96 percent of ˆα
j rsquos
estimated were statistically nonsignificant. Then, a prediction error for each firm j, PE
j t
, was calculated for a period around the date of the first public announcement of the merger using the equation PE
j t
= R
j t
−β
j
× R
mt
, where the prediction error, PE, was the difference between actual stock price and the ex-
pected stock price without merger. The period during which base values for α
j
and β
j
are calculated is excluded. The date of first public announcement is identical to the press date of
Asquith 1983 and is considered to be the first day that a merger rumor, discussion, tender offer, proposal, agreement or understanding appears in the Wall Street Journal.
Estimates of the value created, or the percent change in bidder equity and target equity due to merger, were obtained by summing the PE
j t
abnormal change in the rate of return for stock j on day t for the acquiring firm and the target firm for four different intervals around
the merger announcement date: i the day before and the day of the announcement t = −1, 0, ii the day before and the day after the announcement t = −1, 1, iii 10 days before
and 10 days after announcement t = −10, 10, and iv 10 days before and 25 days after announcement t = −10, 25. Multiple time periods were tested in order to capture any pre
374 L.M. Shelton J. of Economic Behavior Org. 41 2000 363–383
and post announcement gains. Bradley 1980 and Keown and Pinkerton 1981 show that a significant amount of the target firm’s gain can occur before the merger announcement date.
3.3. Independent variables Rivalry. This dummy variable equals one if more than one firm was bidding for the target,
and is zero otherwise. Strategic fit. This variable measures the percentage of related strategic fits in the merger,
which were determined by pairwise comparisons of line of business data for each business unit of the target and bidder as in Shelton 1988. The relationship between each pair
of business units was classified into one of two possible strategic categories: related or unrelated. Businesses units were considered related if they met one of the following criteria:
1. target and bidder business units in the same business; 2. target unit enables bidder to integrate forward or backward, expand product lines or
otherwise consolidate the bidder’s market position; 3. target unit enables bidder to enter new but related markets.
The unrelated category is self-explanatory. If any two target and bidder businesses are considered related using the above criteria, then the percentage of sales for each of the
two related units together are multiplied together to yield a percentage of the combined sales of the mergeri.e. 20 percent × 30 percent = 6 percent. The percentages of the com-
bined sales for each related target — bidder business pair are then summed to yield a total percentage of related combined sales for the merger ranging from 0 to 100 percent. This
procedure provides a continuous measure of relatedness, as opposed to a simple dummy variable.
Merger cycle. For each merger, this variable denotes the number of mergers which oc- curred during the announcement year over 1 MM1000. High values of this variable indicate
that the merger occurred during a peak in the merger cycle, while low values indicate that it occurred during a trough.
Merger cyclestrategic fit. The interaction between merger cycle and strategic fit is mea- sured by multiplying the percentage of related strategic fits by the number of mergers during
the announcement year divided by 1000. Rivalrystrategic fit. The interaction between rivalry and strategic fit is measured by
multiplying the rivalry dummy variable by the percentage of related strategic fits. Merger cyclerivalry. The interaction between rivalry and merger cycle is measured by
multiplying the rivalry dummy variable by the number of mergers during the announcement year divided by 1000.
Institutional holdings. This variable measures institutional holdings divided by total com- mon stock.
Relative size. The relative size of the target was measured by dividing target firm sales by bidder firm sales.
Regulation. This dummy variable equals one if the merger occurred after October 1969 when the Williams Act went into effect, and is zero otherwise.
Equity financing. This dummy variable equals one if the merger was financed by stock exchange, and is zero if the merger was financed by cash.
L.M. Shelton J. of Economic Behavior Org. 41 2000 363–383 375
4. Results
